Workplace drug-testing programs in the United States have proliferated over the past 10-15 years. It appears that many companies see drug-testing programs as an effective policy for deterring illicit drug use among their employees. However, there is little empirical research that has examined the relationship between drug-testing programs and worker drug use. Using data from the 1994 National Household Survey on Drug Abuse (NHSDA), this paper examines the associations among drug use, drug-testing programs, and perceptions of programs. The results suggest that weekly users of marijuana or cocaine are less likely than others to work for companies that have drug-testing programs. In general, users of marijuana and cocaine are also less likely to want to work for such companies. These findings suggest a deterrent effect of drug-testing programs, although the mechanisms underlying this deterrent effect remain unknown.
Workplace drug-testing programs have proliferated over the past decade or so. Annual surveys conducted by the American Management Association (AMA) indicate that the proportion of its constituent companies that test employees for drugs increased from 22% in 1987 to over 81% in 1997 (AMA, 1997). National surveys conducted by the U.S. Bureau of Labor Statistics and the Research Triangle Institute show that the percentage of U.S. firms with drug-testing programs rose from about 20% in 1988 to over 48% in 1993 (Hartwell, Steele, French, & Rodman, 1996; Hayghe, 1991). Although some evidence indicates that the positive rate of these drug tests has gone down (AMA, 1997; SmithKline Beecham Clinical Laboratories, 1997), and a number of indicators show that illicit drug use among U.S. adults in general (Substance Abuse and Mental Health Services Administration, 1998) and U.S. workers in particular (Substance Abuse and Mental Health Services Administration, 1997) has decreased substantially, the prevalence of workplace drugtesting programs has continued to rise. However, it is important to note that while larger companies, such as those represented by the AMA, have turned increasingly to drug testing as a response to perceived worker drug use, there is evidence that many smaller companies have turned away from drug testing (Macdonald & Wells, 1994, citing Cornell/ Smithers, 1992).
Nevertheless it appears that many U.S. companies see drugtesting programs as part of an effective policy for deterring illicit drug use among workers. The deterrent effect may be general in terms of preventing drug users from applying for jobs, or it may be specific by minimizing drug use among current employees. Some might argue that merely comparing aggregate time series of the prevalence of testing programs and the rate of positive test results suggests that the deterrent effect of drug-testing programs is a reality, especially since higher prevalence of testing has accompanied lower rates of positive test results. However, it is also plausible that the reduction in positive drug tests simply reflects a more general societal trend: adults in the U.S., whether employed or not, are less likely to use illicit drugs compared with earlier periods (Horgan, 1990; Substance Abuse and Mental Health Services Administration, 1997, 1998).
Irrespective of the trends, drug testing is part of the general labor landscape in the U.S. and in many other countries (Blackwell, 1994; Hartwell et al., 1996; International Labour Organization, 1994; Macdonald, 1997; Schottenfeld, 1989). But there have been few studies that have provided an empirical examination, directly or indirectly, of the association between worker drug use and workplace drug testing. A number of studies have provided estimates of the percentage of companies that have testing programs (e.g., Hartwell et al., 1996), a few projects have assessed employee attitudes toward drug testing (e.g., Bennett, Blum & Roman, 1994; Blackwell & Grasmick, 1997; Kravitz & Brock, 1997; LeRoy, 1991), and research has examined the link between drug use and work-related outcomes such as accidents, job turnover, and work performance (e. …